Ethan Benson: > > > .... BenH has been talking about throwing away quik's second > > stage code and making yaboot compatible with Oldworld OF and quik's > > first stage. this way there would effectivly be only one codebase for > > both oldworld and newworld bootloaders. only the installer and first > > stage loader would be different. > > > > for oldworld quik's first stage powerpc asm bootblock would be used, > > and its bootblock installer /sbin/quik would be used to save a > > blocklist of the elfextracted yaboot.b. > > > > on newworld you would of course use ybin and its Forth first stage > > loader installed onto a dedicated bootstrap partition. > > > > again i reiterate porting yaboot to oldworld will not make oldworld > > magically as easy to OF boot as a newworld. Oldworld OF is pure crap > > and any peice of software is going to have a helluva time functioning > > correctly on it. some of those machines can't even read from the hard > > disk which you have to fix before the bootloader even enters into it > > (since OF must read the disk to load the bootloader). > > > > BenH myself and a few others are interested in getting OF booting > > cleaned up more because simply put, BootX/miboot are kludges. period, > > macos based booting is fundementally flawed and it would be good for > > all concerned if that would just go away. > > > > ...we need to get him a NuBus/PDS video card which works with current > > kernels > > or an Apple HDI-45 video adapter cable. >
On Sun, May 13, 2001 at 05:11:48PM -0400, [EMAIL PROTECTED] wrote: > My offer of (an Apple HDI-45 video adapter cable) or (MicroConversions > 2124NBII 24-bit NuBus Card, AppleDB15 female) still stands. > > We desperately need to condense the 50+ ways to boot your DebianPowerMac. > The current hodgepodge of incomplete and anectdotal reports on the mailing > list is > too disjointed to allow someone to author a comprehensive HOWTO. > It is currently impossible to create one > without a firm understanding of each unique instance, > and its current status against the various Debian releases. > The pool of potential authors is therefore very, very small. > > It would be a shame to let someone waste their time pondering & composing > all the possible > vagarities of a "kludge" or two when the actual goal is in a different > paradigm. > The problem is there isn't anyone that has worked with all of the various models with linux. What we really need is a way to verify each report for which set of machines. In some cases there may have been past OF patching, and that needs to be considered. If a howto is written, it needs to have specific notes on each tested model/OF combination. We'll need to make sure that they've reset their Open Firmware, and start from a "badblocks -w" disk. Also, we need to choose a way to test the memory in the machines. Memtest only works on mlock-able memory, and that may not be acceptable. Let's start a little verification toolset, and we can get some reliable results. Also, /proc/cpuinfo should be included in any reports. Mike
